refactor(test): keep coverage at behavioral boundaries

[morluto]

Description

Continue the test-debt sweep by assigning tests to the narrowest behavioral owner and removing coverage that only mirrored implementation details, private listener/timer state, type structure, or stronger tests elsewhere.

This change:

• colocates pure domain, adapter, and recording-port service tests with their production owners;
• keeps real filesystem, provider, composition, acceptance, conformance, and public MCP/CLI checks in their behavioral lanes;
• replaces broad managed-provider and error suites with focused owner tests plus retained boundary and conformance coverage;
• consolidates repeated BinarySession targets, providers, deferred work, and controllable clients into one shared fixture;
• keeps the public process-capture example in production while excluding test-only fixture helpers from the package build;
• shares the immutable MCP module graph across boundary files while each test still creates and closes its own session.

The sweep does not change tool contracts, provider authority, persistence semantics, or authorization behavior. Domain schema refactoring is intentionally excluded and tracked separately in #473.

Performance

On the same local Linux checkout:

• MCP boundary lane: 118.32s to 40.98s for the same 49 files / 132 tests after removing per-file module isolation.
• Composition lane: 18.25s for 23 files / 66 tests to 16.23s for 21 files / 58 retained behavioral tests after moving filesystem boundaries and removing duplication.

Type of Change

• Bug fix
• New feature
• Documentation update
• Performance improvement
• Refactoring (no functional change)
• Tool contract change (MCP schema modification)

Testing Performed

• Unit tests added or updated
• npm run check:pr — typecheck, lint, formatting, Knip, metadata, build, 344 test files / 1,697 tests, TypeDoc, and generated-document checks passed
• npm pack --dry-run — 738 files, 1.1 MB package
• Pre-push npm run check:fast — typecheck and lint passed
• npm run verify:package — package construction and startup checks ran until the Linux Hopper platform probe; this host lacks Xvfb, so the verifier stopped with missing_xvfb
• Real Hopper, Ghidra, browser, replay, and managed-provider verification performed

The exact-diff review was stopped at the explicit publish request after one clean bounded pass; no actionable finding had been reported.

Tool Contract Impact

None. The canonical tool inventory and generated catalog are unchanged.

Suggested review order

1. vitest.config.ts, docs/testing.md, and the process-global lane classifier.
2. Source-owned fixtures and focused owner tests under src/domain, src/application, and provider directories.
3. Retained boundary/composition/conformance tests and BinarySession fixture consolidation.

Related Issues

No linked issue.

[cursor]

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.

[morluto]

@codex

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Another round soon, please!

Reviewed commit: 42654375f0

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".